Nav1.1 Modulation by a Novel Triazole Compound AttenuatesEpileptic Seizures in RodentsJohn Gilchrist,† Stacey Dutton,‡ Marcelo Diaz-Bustamante,§ Annie McPherson,‡ Nicolas Olivares,§

Jeet Kalia,∥ Andrew Escayg,*,‡ and Frank Bosmans*,†,⊥

†Department of Physiology, Johns Hopkins University, School of Medicine, Baltimore, Maryland 21205, United States‡Department of Human Genetics, Emory University, School of Medicine, Atlanta, Georgia 30022, United States§Lieber Institute for Brain Development, Johns Hopkins University, School of Medicine, Baltimore, Maryland 21205, United States∥Indian Institute of Science Education and Research Pune, Pune, Maharashtra 411 008, India⊥Solomon H. Snyder Department of Neuroscience, Johns Hopkins University, School of Medicine, Baltimore, Maryland 21205,United States

*S Supporting Information

ABSTRACT: Here, we report the discovery of a novel anticonvulsant drugwith a molecular organization based on the unique scaffold of rufinamide, ananti-epileptic compound used in a clinical setting to treat severe epilepsydisorders such as Lennox-Gastaut syndrome. Although accumulating evidencesupports a working mechanism through voltage-gated sodium (Nav) channels,we found that a clinically relevant rufinamide concentration inhibits human(h)Nav1.1 activation, a distinct working mechanism among anticonvulsants anda feature worth exploring for treating a growing number of debilitatingdisorders involving hNav1.1. Subsequent structure−activity relationshipexperiments with related N-benzyl triazole compounds on four brain hNavchannel isoforms revealed a novel drug variant that (1) shifts hNav1.1 openingto more depolarized voltages without further alterations in the gating propertiesof hNav1.1, hNav1.2, hNav1.3, and hNav1.6; (2) increases the threshold toaction potential initiation in hippocampal neurons; and (3) greatly reduces the frequency of seizures in three animal models.Altogether, our results provide novel molecular insights into the rational development of Nav channel-targeting molecules basedon the unique rufinamide scaffold, an outcome that may be exploited to design drugs for treating disorders involving particularNav channel isoforms while limiting adverse effects.

Severe epileptic disorders such as Lennox-Gastaut syndrome(LGS) are commonly associated with various types ofseizures and cognitive dysfunction that persist into adulthood.1

As a result, these patients suffer from varying degrees oflearning disabilities, psychiatric disorders, and behavioralabnormalities that can drastically affect their social integration.1

Diagnosis and subsequent management of LGS are challenging,relying mostly on the expertise of the physician to interpret anintricate amalgamation of clinical and EEG abnormalities.2

Although limited clinical trials suggest that a small subset ofavailable anti-epileptic drugs (AEDs) such as lamotrigine,valproic acid, topiramate, felbamate, and clobazam may be usedto manage the multiple seizure types associated with LGS, theirefficacy is often inadequate, and perilous adverse effects mayoccur.2 The relatively new drug rufinamide, a compound withorphan drug status in the United States and marketingauthorization in Europe, displays a broad spectrum of anti-epileptic activity, and clinical trials have shown long-termbeneficial effects and good tolerability when rufinamide is usedas adjunctive therapy in children and adults suffering fromLGS.3,4 As such, rufinamide is poised to replace the more

adverse effect-prone classical AEDs as a valuable treatment forLGS5 as well as partial3 or refractory focal6 seizures.Interestingly, the hydrophobic triazole-derived molecularstructure of rufinamide is unlike that of any other anti-epilepticcompound;4 nonetheless, the drug is absorbed extensivelywhen taken orally, and dosages of up to 3200 mg/day arecommon.4,7,8

Given their unique role in electrical signaling9 andsubsequent implications in various epilepsy disorders,10−14

voltage-gated sodium (Nav) channels within the central nervoussystem (CNS) may be influenced by rufinamide;15,16 however,it is unclear whether modulating a particular isoform constitutesthe primary mode of action. Of the nine Nav channel isoforms(Nav1.1−Nav1.9) identified in humans, four are expressedpredominantly in the CNS [Nav1.1 (SCN1A), Nav1.2(SCN2A), Nav1.3 (SCN3A), and Nav1.6 (SCN8A)].

9,17

Supporting their physiological importance, mutations in

Received: September 17, 2013Accepted: March 17, 2014Published: March 17, 2014

Articles

pubs.acs.org/acschemicalbiology

© 2014 American Chemical Society 1204 dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−1212

Open Access on 03/17/2015

pubs.acs.org/acschemicalbiologyhttp://pubs.acs.org/page/policy/authorchoice/index.html

Nav1.1,12,18,19 Nav1.2,

13 and Nav1.620,21 have been linked to

various epilepsy phenotypes, whereas Nav1.3 is believed to beinvolved also in nociception after channel upregulation due tospinal cord injury.22−24 Consequently, we set out to investigatewhether rufinamide influences the functional properties of oneor more of these four Nav channel isoforms. To this end, wetransiently expressed human (h) hNav1.1, hNav1.2, hNav1.3,and hNav1.6 in a robust heterologous expression system andfound that a clinically relevant rufinamide concentrationprimarily influences the voltage dependence of hNav1.1activation and availability of hNav1.6 channels, whereas therecovery from fast inactivation of hNav1.1, hNav1.2, hNav1.3,and hNav1.6 is only slightly altered. Subsequent experimentswith structurally related compounds reveal that chemicalderivatization at particular positions on the triazole pharmaco-phore may be exploited to obtain specific inhibition of hNav1.1activation, thereby increasing seizure thresholds in rodentmodels of epilepsy.

■ RESULTSExpressing Human Nav Channel Isoforms in Xenopus

Oocytes. To identify the molecular target of rufinamide in thebrain, we examined potential drug-induced alterations ofhuman neuronal Nav channel opening and closing (i.e., gating).Although stable cell lines of hNav1.1, hNav1.2, hNav1.3, andhNav1.6 have been reported,

29−32 DNA rearrangement eventsand low protein levels have hampered in-depth experimentswith these genes in Xenopus laevis oocytes, a transient androbust heterologous expression system widely used to addressfundamental questions about the gating mechanisms andpharmacological sensitivities of individual ion channel isoforms.By combining careful full-length clone sequencing withtransformations of cDNA modified with an ER forwardingmotif25 and the rNav1.2a 3′UTR region into low-copyEscherichia coli variants (CopyCutter EPI400), we were ableto consistently obtain robust ionic currents for all four humanNav channel isoforms in Xenopus oocytes (Figure S1 of theSupporting Information). Examination of the G−V relation-ships for hNav1.1, hNav1.2, hNav1.3, and hNav1.6 (Figure S1 ofthe Supporting Information) reveals that the midpoints (V1/2)for channel activation are −37.8, −31.4, −25.4, and −36.7 mV,respectively (Table 1). Although the same relative order ofmidpoints is essentially observed in mammalian cell record-ings,29−32 their absolute values as well as those from channel

availability and recovery from inactivation measurements(Figure S1 of the Supporting Information and Table 1) differnoticeably, yet this observation is not surprising considering thevariability in lipid membrane, glycosylation, and auxiliarysubunit composition between diverse cell types. Next, weapplied rufinamide to hNav1.1−hNav1.3 and hNav1.6 todetermine whether the drug alters channel gating.

The Gating Process of Human Nav Channels Is Alteredby Rufinamide. Rufinamide, or 1-[(2,6-difluorophenyl)-methyl]-1H-1,2,3-triazole-4-carboxamide,4 is slightly soluble inwater but dissolves readily in dimethyl sulfoxide (DMSO) atconcentrations of up to 9 mg mL−1 (Figure 1A). As such,patients are given the drug either as a tablet formulation(Banzel) or as an oral suspension (Inovelon), preferably to betaken with food.7 On the basis of these properties, we dissolvedrufinamide in DMSO and diluted this stock solution withappropriate recording media to a final concentration of 100μM, thereby approximating clinically observed serum quanti-ties.4,8,33 Subsequently, Nav channel-expressing cells wereexposed to 100 μM rufinamide in a solution containing atmost 1% DMSO. Control experiments without the drugrevealed that the gating properties of hNav1.1, hNav1.2,hNav1.3, and hNav1.6 are not significantly affected uponexposure to 1% DMSO (Figure S2 of the SupportingInformation and Table 1). Hereafter, we will consider thegating parameters from DMSO-treated Nav channel isoforms tobe our control values.After incubating oocytes with 100 μM rufinamide, we

observed an ∼8 mV depolarizing shift in the G−V relationshipof hNav1.1, whereas the activation of hNav1.2, hNav1.3, andhNav1.6 is not inhibited (Figure 1 and Table 1). In contrast,steady-state inactivation of hNav1.1, hNav1.2, and hNav1.3 isnot influenced; however, 100 μM rufinamide does alter themidpoint of hNav1.6 channel availability by approximately +5mV (Figure 1 and Table 1). Moreover, recovery from fastinactivation slows for all tested Nav channel isoforms (Figure 1and Table 1), and although these effects are subtle, thecombination with a substantial shift in hNav1.1 activationvoltage may help explain a decrease in neuronal excitability afteradministration of the drug.15 Altogether, our experiments withfour neuronal Nav channel isoforms suggest that rufinamideprimarily influences hNav1.1 and hNav1.6 function, anobservation that supports a role of these particular Nav channelvariants in epilepsy syndromes.10,12−14,18,20,21 It is worth noting

Table 1. Effects of 100 μM Rufinamide and Its Derivatives on Human Nav Channel Isoformsa

hNav1.1 hNav1.2 hNav1.3 hNav1.6

control activation (V1/2) (mV) −37.8 ± 1.7 −31.4 ± 1.5 −25.4 ± 1.4 −29.0 ± 2.2inactivation (V1/2) (mV) −52.9 ± 1.0 −57.1 ± 0.3 −52.2 ± 0.7 −64.8 ± 1.2recovery (T) (ms) 3.4 ± 0.2 6.6 ± 0.5 5.6 ± 0.3 4.4 ± 0.3

DMSO activation (V1/2) (mV) −38.2 ± 1.9 −31.7 ± 1.2 −24.7 ± 1.6 −26.7 ± 1.1inactivation (V1/2) (mV) −49.4 ± 0.3 −59.7 ± 0.4 −52.8 ± 0.9 −65.1 ± 0.8recovery (T) (ms) 3.8 ± 0.5 7.1 ± 0.4 6.5 ± 0.3 5.0 ± 0.4

rufinamide activation (V1/2) (mV) −30.6 ± 0.9* −29.5 ± 1.2 −24.5 ± 2.4 −28.1 ± 0.8inactivation (V1/2) (mV) −47.8 ± 1.6 −57.7 ± 0.5 −52.6 ± 1.6 −60.0 ± 0.7*recovery (T) (ms) 5.1 ± 0.2* 10.6 ± 0.3* 9.2 ± 0.5* 7.4 ± 0.3*

compound A compound B compound C compound D

hNav1.1 activation (V1/2) (mV) −30.8 ± 1.0* −27.4 ± 1.0* −38.4 ± 1.3 −32.5 ± 2.3inactivation (V1/2) (mV) −50.5 ± 1.2 −49.4 ± 0.3 −49.2 ± 0.6 −49.8 ± 1.6recovery (T) (ms) 4.1 ± 0.5 3.9 ± 0.1 3.1 ± 0.1 3.9 ± 0.3

aDrug values are compared to DMSO values for statistical comparison, which is considered significant if the result of the Student’s t test indicated a pof

that rufinamide may alter hNav1.1 activation by influencingvoltage-sensor activation or by slowing subsequent gatingtransitions,34 a distinct working mechanism among the classicanticonvulsant drugs because these compounds are thought toexert their effect by (1) occluding the pore to prevent sodiumion flow or (2) interacting with the inactivated state to decreasethe size of the pool of channels available for opening.33,35−37

Next, we wanted to explore whether particular structural

features of the rufinamide molecule can enhance hNav1.1selectivity.

Molecular Modifications of Rufinamide Enhance theEffects on hNav1.1. The triazole-derived molecular organ-ization of rufinamide is unique among anti-epileptic com-pounds.4 To examine which features of the N-benzyltriazolecore are important for its inhibitory action on hNav1.1, weperformed a structure−activity relationship (SAR) study usingfour structurally related compounds with unknown function-ality: 1-[(3-fluorophenyl)methyl]-1H-1,2,3-triazole-4-carboxa-mide-5-amine (compound A), 1-(phenylmethyl)-1H-1,2,3-triazole-4-carboxamide-5-methyl (compound B), 1-[(3-fluorophenyl)methyl]-1H-1,2,3-triazole-4-hydroxymethyl(compound C), and 1-[(4-fluorophenyl)methyl]-1H-1,2,3-triazole-4-carboxamide-5-amine (compound D). We foundthat compounds A and B inhibit hNav1.1 opening (Figure 2and Table 1) and compound B produces the largestdepolarizing shift in channel activation voltage (approximately

Figure 1. Effect of 100 μM rufinamide on human Nav channelisoforms involved in epilepsy. (A) Molecular organization ofrufinamide consisting of the 1,2,3-triazole ring connecting the 2,6-difluorophenyl and the carboxamide. (B−E) The left column showsG/Gmax and I/Imax relationships, whereas the right column displays therecovery from fast inactivation. The figure shows that rufinamideinhibits hNav1.1 opening whereas the effect on hNav1.6 is notsignificant compared to the effect of DMSO treatment of thisparticular isoform (see Table 1). Furthermore, the recovery from fastinactivation slows for the four Nav channel isoforms tested here. Alldata are shown before (green, DMSO control) and after (red) additionof 100 μM rufinamide to hNav1.1 (B), hNav1.2 (C), hNav1.3 (D), andhNav1.6 (E). All fit values are listed in Table 1. n = 5−8, and error barsrepresent the standard error of the mean.

Figure 2. Effect of rufinamide derivatives on hNav1.1. (A−D) The leftcolumn shows G/Gmax relationships before (green, DMSO control)and after (red) addition of 100 μM rufinamide derivatives fitted withthe Boltzmann equation. Compounds A and B clearly inhibit hNav1.1opening, whereas compound C no longer affects hNav1.1 opening(compound D is not significant when considering p < 0.005). Fitvalues are listed in Table 1. n = 5−8, and error bars represent thestandard error of the mean. The right column displays the molecularorganization of the four tested derivatives (compounds A−D).

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121206

+11 mV). Strikingly, neither of these compounds influenceshNav1.1 steady-state inactivation, recovery from inactivation, orthe persistent current38 (Figures S3 and S4A,B of theSupporting Information and Table 1), suggesting a workingmechanism geared toward stabilizing a closed state of thisparticular Nav channel isoform. In contrast to the effect ofrufinamide, application of 100 μM compound B selectivelymodulates hNav1.1 activation whereas the tested gatingparameters of hNav1.2, hNav1.3, and hNav1.6 are unaltered(Figure S5 and Table S1 of the Supporting Information andTable 1). Moreover, our results suggest that rufinamide andcompound B do not influence the entry into or voltagedependence of hNav1.1 slow inactivation (Figure S6 and TableS2 of the Supporting Information).39 Finally, rufinamide andcompound B (100 μM) do not activate the α1/β2/γ2 GABAAreceptor (Figure S4C of the Supporting Information), asubtype ubiquitously found in GABAergic neurons that istargeted by psychoactive drugs such as zolpidem andbenzodiazepines.40 Also, neither drug inhibits hERG,27 amember of the cardiac potassium channel family and anFDA-mandated screening target for potential off-target drugeffects (Figure S7 of the Supporting Information).One of the most captivating results from our SAR studies is

that compound B inhibits hNav1.1 more efficaciously thanrufinamide (Figure 2 and Table 1). The identification of such amolecule is particularly exciting as it underlines the broad scopeof rufinamide scaffold-based drug development for treatinghNav1.1-related disorders. Upon comparison of the structuresof rufinamide and compound B, two differences stand out. Onone hand, rufinamide has two electron-withdrawing fluorosubstituents on the phenyl group whereas compound B hasnone, thereby rendering the phenyl group of this moleculemore electron-rich. On the other hand, there is an additionalmethyl substituent on the triazole ring in compound B,resulting in an increased level of hydrophobic characterincreased compared to that of rufinamide. Another importantobservation from our experiments is that compound D hasdiminished activity whereas compound C no longer influenceshNav1.1 gating at all (Figure 2c). The most significantstructural feature of compound C is the presence of amethylene hydroxyl (-CH2OH) group on its triazole moiety;an amide (-CONH2) substituent is found in the othercompounds. To determine whether compound B is physiolog-ically active, we next examined the activity of this molecule inhippocampal neurons.Compound B Increases the Action Potential Thresh-

old in Hippocampal Neurons. The Nav1.1 distribution isstrikingly similar in human and rodent brains.41 Moreover, 99%of the amino acids that make up Nav1.1 are conserved amonghumans, rats, and mice (ClustalΩ42). As such, we expect theinhibitory effect of compound B to increase the action potentialthreshold in Nav1.1-expressing rat hippocampal neurons. Totest this hypothesis, we applied a physiological solutioncontaining 100 μM compound B to cultured hippocampalneurons and found that a current injection of 80 pA elicits anaction potential (Figure 3a). In contrast, hippocampal neuronsunder control conditions require a current injection of only 20pA to start generating action potentials (Figure 3A). Whenassessing average firing thresholds by calculating the potentialat which the rate of rise crosses 40 V/s, we observe a value of−43.3 ± 2.4 mV for control cells, whereas a threshold of −27.8± 1.6 mV is noted when cells are treated with 100 μMcompound B (Figure 3B). Altogether, these results suggest an

ability of compound B to reduce the level of action potentialfiring in hippocampal neurons by shifting the firing threshold inthe depolarizing direction, possibly by deferring Nav1.1 channelactivation to more positive membrane voltages (Figure 2B,Figure S5 of the Supporting Information, and Table 1).Subsequently, we examined the biological activity of thismolecule in three established mouse epilepsy models.

Compound B Is Effective in Rodent Seizure InductionParadigms. First, we tested compound B in the picrotoxinseizure induction assay, a widely used epilepsy model based oninhibiting GABA-mediated synaptic transmission. On the basisof the effective dose of rufinamide reported previously,5 wetested compound B at a dose of 75 mg/kg. Even though theforelimb clonus (FC) tends to occur later in the compound B-treated mice, we observe no statistically significant differencesbetween vehicle- and compound B-treated mice in their averagelatencies to the first myoclonic jerk (MJ) (p = 0.4) and FC (p =0.3) (Figure 4). However, the average latency to the firstgeneralized tonic-clonic seizure (GTCS) following treatmentwith compound B is significantly increased by 41% (1693 ±315 s; p = 0.03) compared to that of vehicle-treated mice (1006± 106 s). There is no difference in the frequency of thebehavioral seizures between the treatment groups.Because one assay may not accurately reflect the ability of a

drug to increase seizure resistance, we next investigatedcompound B efficacy in the fluorothyl paradigm, an establishedseizure induction model involving exposure of the animal to thevolatile proconvulsant bis-2,2,2-trifluoroethyl ether.43 Althoughthe average latency to the MJ is comparable between vehicle-and compound B-treated mice (p = 0.6), the average latenciesto the first GTCS and GTCS with hind limb extension (GTCS+) are increased by 43 and 56%, respectively (p < 0.0001; 652± 48 s for GTCS and 1155 ± 89 s for GTCS+), compared tothose of the vehicle-treated group (281 ± 19 and 652 ± 38 s,respectively) (Figure 5). These results clearly demonstrate thatcompound B successfully increases the average latency tofluorothyl-induced GTCS and GTCS+.Finally, we examined whether compound B is capable of

reducing the occurrence and severity of seizures observed in the6 Hz seizure induction paradigm, a model with a proven trackrecord in drug development.44 At a current intensity of 17 mA,92% (n = 22 out of 24) of vehicle-treated mice displaybehavioral seizures and the majority of the observed seizurescan be characterized as forelimb clonus (Racine score of 2) orrearing and falling (Racine score of 3) (Figure 5). In sharp

Figure 3. Effect of compound B on rat hippocampal neurons. (A)Representative recordings of cells treated with DMSO-containingvehicle (left) vs cells treated with 100 μM compound B (right). Theinset shows the current clamp protocol of a series of 10 pAdepolarizing current injections in which red and green indicate thecurrent needed to elicit an action potential under control conditions(20 pA) and that of the treated cells (80 pA), respectively. (B) Averageaction potential thresholds are significantly higher in cells treated withcompound B [−27.8 ± 1.6 mV (○); n = 7] than in cells treated withthe DMSO-containing vehicle [−43.3 ± 2.4 mV (●); n = 7].

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121207

contrast, only 29% (n = 7 out of 24; p < 0.0001) of thecompound B-treated animals exhibit behavioral seizures, whichare also dramatically less severe compared to those seen in thevehicle-treated mice (p < 0.0001). Overall, these resultssubstantiate the notion that it is possible to achieve seizureresistance by modifying the N-benzyltriazole scaffold ofrufinamide to target hNav1.1 activation.

■ DISCUSSIONThe results presented here together with previously reportedobservations15,16,33 suggest that the anti-epileptic drugrufinamide primarily targets the hNav1.1 and hNav1.6 isoformsin the brain (Figure 1 and Table 1). However, it is worthmentioning that additional mechanisms of action cannot beruled out. For example, in an effort to find drugs forneuropathic pain, researchers found that rufinamide stabilizesthe inactivated state of the Nav1.7 isoform that is typicallyfound in the peripheral nervous system.33,45 Although this effectis rather small compared to that on hNav1.1 activation voltage,it may contribute to a beneficial effect of rufinamide in patientswith febrile seizures.46 Because anomalous hNav1.1 behaviorhas been implicated in LGS,1,2,47−49 the therapeutic effects ofrufinamide may indeed stem from its ability to shift the voltagedependence of activation of hNav1.1 channels to more positivepotentials (Figure 1 and Table 1), a notion worth exploring fortreating disorders involving hNav1.1 gain-of-function mutations

such as GEFS+,10,14,18,50−53 migraine,54,55 and post-traumaticstress disorder.56 However, rufinamide may be contraindicatedwith patients in whom Nav1.1 function is already impaired. Forinstance, Nav1.1 deletion in mouse hippocampal and corticalinterneurons decreases the level of sustained high-frequencyfiring of action potentials by reducing the overall sodiumcurrent without altering its voltage-dependent features.28,57,58

Subsequently, loss of Nav1.1 function may lead to hyper-excitability and epileptic seizures in patients with Dravetsyndrome58 and autism.59,60 Although more experiments areneeded to explain this phenomenon, it is worth noting thatNav1.1 deletion in mice causes Nav1.3 to be upregulated,

58 oneof several observations that highlight a delicate balance betweenvarious signaling components involved in cellular excitation andinhibition that, in turn, complicates the translation of inhibitoryanticonvulsant efficacy from heterologous expression systems toan intricate neuronal circuit.11,15,35,36,61

Intrigued by the unique molecular organization of rufinamide(Figure 1A), we examined which features of the N-benzyltriazole core contribute most to its inhibitory effect onhNav1.1. As a result, three important features that need to beconsidered for hNav1.1 drug development using the N-benzyltriazole scaffold emerge from our SAR study. First,tweaking the electron density of the phenyl ring by removing orrepositioning electron-withdrawing fluoro groups is a strategyworth exploring. Second, increasing triazole ring hydro-phobicity by introducing alkyl substituents may give rise tomolecules with better hNav1.1 efficacy. Third, an amidesubstituent on the triazole ring plays an important role inimparting hNav1.1 selectivity to rufinamide and its analogues.Considering these results, it will be very insightful to synthesizea series of rufinamide analogues bearing different alkyl chains incombination with varying electron density-altering substituentson the phenyl group and then test their efficacy as anti-epilepticagents. Subsequently, we identified a molecule with theproperties mentioned above that does not influence hNav1.2,hNav1.3, or hNav1.6 activation or steady-state inactivation(compound B in Figure 2B) and examined its effect on Nav1.1-expressing cultured hippocampal neurons (Figure 3). Asanticipated when shifting the Nav channel activation voltage,we observed a significant increase in the action potentialthreshold (∼16 mV) in the presence of 100 μM compound B(Figure 3), an exciting result that motivated us to translate ourfindings to three rodent models of epilepsy. Consistent with a

Figure 4. Compound B increases latencies to picrotoxin- andfluorothyl-induced generalized seizures. (A) Picrotoxin (10 mg/kg)was administered to vehicle-treated (black) and compound B-treated(red) male Crl:CF1 mice (75 mg/kg) 15 min prior to seizureinduction. The average latencies in seconds to the first myoclonic jerk(MJ), forelimb clonus (FC), and generalized tonic-clonic seizure(GTCS) are compared. Compound B increases the average latency tothe first GTCS. *p < 0.05. Error bars represent the standard error ofthe mean. (B) The latencies to the MJ, GTCS, and GTCS with hindlimb extension (GTCS+) in the fluorothyl model are comparedbetween vehicle-treated (black) and compound B-treated (red) maleCrl:CF1 mice (75 mg/kg). The average latencies in seconds to theGTCS and GTCS+ are significantly longer in the compound B-treatedmice. ***p < 0.001. Error bars represent the standard error of themean.

Figure 5. Compound B reduces the severity of seizures induced by the6 Hz psychomotor paradigm. A current intensity of 17 mA was used toinduce partial seizures in vehicle-treated (black) and compound B-treated (red) male Crl:CF1 mice (75 mg/kg). Following treatmentwith compound B, there is a 63% reduction in the number of miceexhibiting seizures (A; p < 0.0001). In addition, the seizures that wereobserved in the compound B-treated mice are typically less severe thanthose seen in the vehicle-treated mice (black) (B; p < 0.0001). Errorbars represent the standard error of the mean.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121208

role for Nav1.1 in prompting neuronal firing, we found thatcompound B is capable of increasing the average latency to theGTCS in the picrotoxin (Figure 4) and fluorothyl paradigms(Figure 5). In addition, compound B reduces seizureoccurrence and severity in the 6 Hz seizure induction assay(Figure 5), a model previously employed to successfullyidentify anticonvulsant drugs such as tiagabine,62 vigabatrin,62

and lacosamide.63 Overall, these results demonstrate thatmodifying the unique N-benzyltriazole scaffold of rufinamidecan produce biologically active molecules that target hNav1.1opening, an outcome that may be exploited in the design ofsuitable compounds for investigating the functional role ofhNav1.1 in healthy and diseased tissues or for treating hNav1.1-related gain-of-function disorders such as GEFS+10,14,18,50−53

and migraine.54,55 Consistent with the protection achieved inwild-type mice, compound B in particular may also find use as ageneral anticonvulsant. A more comprehensive interpretation ofour results is that they provide a conceptual example fordesigning drugs that interact with specific Nav channel isoforms,a necessary feature for reducing unwanted adverse effects.

■ METHODSChannel Constructs. Human (h) hNav1.1 (NM_001165963.1),

hNav1.2 (NM_021007.2), hNav1.3 (NM_006922.3), hNav1.6(NM_014191.3), and hβ1 (NM_001037.4) clones were obtainedfrom OriGene Technologies, Inc., and modified with an ER forwardingmotif25 and the rNav1.2a 3′UTR region for expression in Xenopuslaevis oocytes (acquired from Xenopus One). To check for undesiredrearrangement events, the DNA sequence of all constructs wasconfirmed by automated DNA sequencing before further usage. cRNAwas synthesized using T7 polymerase (Life Technologies) after theDNA had been linearized with the appropriate restriction enzymes.Two-Electrode Voltage-Clamp Recording from Xenopus

Oocytes. Channels were expressed in Xenopus oocytes togetherwith β1 in a 1:5 molar ratio, and currents were studied followingincubation for 1−2 days after cRNA injection [incubated at 17 °C in96 mM NaCl, 2 mM KCl, 5 mM HEPES, 1 mM MgCl2, 1.8 mMCaCl2, and 50 μg/mL gentamycin (pH 7.6) with NaOH] using two-electrode voltage-clamp recording techniques (OC-725C, WarnerInstruments) with a 150 μL recording chamber. Heterologous GABAAreceptor and hERG (Kv11.1) manipulations were achieved usingpreviously reported procedures.26,27 Nav channel data were filtered at4 kHz and digitized at 20 kHz using pClamp10 software (MolecularDevices). Microelectrode resistances were 0.5−1 MΩ when thedevices were filled with 3 M KCl. The external recording solutioncontained 100 mM NaCl, 5 mM HEPES, 1 mM MgCl2, and 1.8 mMCaCl2 (pH 7.6) with NaOH. All experiments were performed at roomtemperature (RT) (∼22 °C). Leak and background conductances,identified by blocking Nav channels with tetrodotoxin, were subtractedfor all of the Nav channel-mediated currents. Chemicals andcompounds A−D were obtained from Sigma-Aldrich, Vitasmlabs,and Maybridge (eMolecules).Analysis of Channel Activity and Drug−Channel Interac-

tions. Unless otherwise specified, voltage−activation relationshipswere obtained by measuring peak currents and calculatingconductance (G), and a Boltzmann function was fit to the dataaccording to the equation G/Gmax = [1 + e

−zF(V−V1/2)/RT]−1, where G/Gmax is the normalized conductance, z is the equivalent charge, V1/2 isthe half-activation voltage, F is Faraday’s constant, R is the gasconstant, and T is the temperature in kelvin. Rufinamide andderivatives were dissolved in DMSO (∼5 mg mL−1 stock solution)and ad hoc diluted to a working concentration of 100 μM while takinggreat care not to reach the maximal solubility of the compounds.Xenopus oocytes expressing the various Nav channel isoforms wereincubated in solutions containing 100 μM drug (final DMSOconcentration of ≤1%) for at least 1 h. We did not observe asignificant alteration in drug-induced effects on Nav channels between

1 h and overnight incubation. Differences in reported values were onlyconsidered significant when p < 0.005 (Student’s t test). Off-line dataanalysis and statistical analysis were performed using Clampfit10(Molecular Devices), Excel (Microsoft Office), and Origin 8(OriginLab).

Action Potential Threshold Recordings in HippocampalNeurons. Hippocampal neurons were dissected from Sprague-Dawleyrat embryos on embryonic day 18. Cells were treated with papain(Worthington Biochemical), dissociated with a pipet, and plated overcoverslips coated with collagen (Life Technologies) and poly-D-lysine(Sigma-Aldrich). Astrocyte beds were prepared at a density of 80000cells/mL and cultured in DMEM (Life Technologies) with 10% fetalbovine serum and 6 mM glutamine for 14 days in 5% CO2 at 37 °C.All of the medium was changed every week. Neurons were plated overconfluent astrocyte beds at a density of 150000 cells/mL and culturedfor 3 weeks in neurobasal medium (Life Technologies) supplementedwith B27 and 2 mM glutaMAX. Half of the medium was changedevery 3−4 days. Cells were treated for 3 h with compound B in 0.3%DMSO or a control (medium containing 0.3% DMSO) beforerecording. Coverslips were transferred to a chamber perfused at a rateof 2 mL/min at 32 °C with ACSF containing 140 mM NaCl, 5 mMKCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM HEPES, and 10 mM glucose(pH 7.3 adjusted with NaOH). An Axopatch 200B instrumentequipped with a Digidata 1440 digitizer and pClamp10 software wereused. Whole-cell current clamp recordings were performed usingborosilicate glass pipettes (4−6 MΩ) filled with a solution containing135 mM potassium gluconate, 20 mM KCl, 10 mM HEPES, 4 mMMg-ATP, 0.3 mM Na2GTP, and 0.5 mM EGTA. Action potentialthresholds were measured by a series of 100 ms, 10 pA depolarizingcurrent injections. At the minimal current injection needed to elicit anaction potential, the firing threshold was determined by calculating thepotential at which the rate of rise crossed 40 V/s, using AxoGraph X.The input resistance was measured as the slope of the linear range ofthe current−voltage plot constructed from the steady-state voltageresponse to depolarizing current injections of 5 pA ranging from −40to 0 pA with a duration of 500 ms. All cells measured were in the rangeof 300−500 MΩ, showing no significant difference betweentreatments (435 ± 38 MΩ for the control and 387 ± 32 MΩ forcompound B with n = 5 for both sets of experiments). Data arepresented as means ± the standard error of the mean. For statisticalanalysis, a Student’s t test was performed using GraphPad Prism 6.The Johns Hopkins University Institutional Animal Care and UseCommittee approved all experimental protocols involving rats.

Animal Husbandry and Drug Administration. Male Crl:CF1mice (10−12 weeks old, Charles River) were housed under a 12 hlight−dark cycle with free access to food and water, and allexperiments were conducted between 10 a.m. and 4 p.m. 1-(Phenylmethyl)-1H-1,2,3-triazole-4-carboxamide-5-methyl (com-pound B) (Sigma-Aldrich) was suspended in a 40% solution of 2-hydroxypropyl-β-cyclodextrin (Sigma-Aldrich) in sterile saline (0.9%),and 15 min prior to each seizure induction paradigm, compound B (75mg/kg) was administered via intraperitoneal (ip) injection. Vehicle-treated animals were similarly treated with 40% cyclodextrin andserved as the controls for each experiment. The Emory UniversityInstitutional Animal Care and Use Committee approved allexperimental protocols involving mice.

Seizure Induction Paradigms and Analyses. PicrotoxinSeizure Induction. Fifteen minutes prior to ip injection of picrotoxin(10 mg/kg, Sigma-Aldrich), mice were treated with the vehicle orcompound B (n = 12 per group). Latencies to the first myoclonic jerk,forelimb clonus, and generalized tonic-clonic seizure were comparedbetween compound B- and vehicle-treated mice. The myoclonic jerk ischaracterized by a sudden jerk of the upper body, whereas forelimbclonus involves rapid movement of the forelimbs for at least 5 s.Finally, the generalized tonic-clonic seizure involves the loss ofpostural control with rapid movement of the limbs for at least 5 s.

Fluorothyl Seizure Induction. Latencies to fluorothyl-inducedseizures were determined as previously described.28 Briefly, fluorothyl[2,2,2-trifluoroethyl ether (Sigma-Aldrich)] was dispensed at a rate of20 μL/min into a Plexiglas chamber, and latencies to three behavioral

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121209

phenotypes were compared between vehicle-treated and compound B-treated (75 mg/kg) mice (n = 12 per group): (1) the first myoclonicjerk, (2) a generalized tonic-clonic seizure, and (3) a generalized tonic-clonic seizure with tonic extension. Similar to picrotoxin-mediatedseizure induction, the myoclonic jerk is the first observable behavioralphenotype followed by the generalized tonic-clonic seizure that ischaracterized by a loss of postural control with rapid movement of allfour limbs for at least 5 s. Lastly, the generalized tonic-clonic seizurewith tonic hind limb extension involves the loss of postural control andrapid movement of all four limbs followed by downward extension ofthe limbs parallel to the body.6 Hz Psychomotor Seizure Induction. Topical anesthetic (0.5%

tetracaine hydrochloride ophthalmic solution) was applied to thecornea 30 min before testing. Fifteen minutes prior to seizureinduction, mice were treated with the vehicle or compound B (n = 12per group). Each mouse was manually restrained, and cornealstimulation (0.2 ms pulse, 6 Hz, 3 s) was performed at a currentintensity of 17 mA (ECT Unit 57800, Ugo Basile, Comerio, Italy).The mice were observed for behavioral responses immediatelyfollowing stimulation, and seizures were scored on the basis of amodified Racine scale: 0, no behavior; 1, staring; 2, forelimb clonus; 3,rearing and falling. The mice were randomized into two groups, and inthe first trial, one group received the vehicle and the other compoundB. The second trial was performed 1 week later, with each groupreceiving the opposite treatment. The resulting data were combinedbecause no statistically significant differences were observed betweenthe two trials.Statistical Analysis. A Student’s t test was used to identify

statistically significant differences in latencies between vehicle- andcompound B-treated groups for each seizure phenotype followingadministration of fluorothyl and picrotoxin. A Fisher Exact test wasapplied to determine differences in the percent of vehicle- andcompound B-treated mice exhibiting seizures in the 6 Hz seizureinduction paradigm, and a Wilcoxon matched-pairs signed rank testdetermined the differences in seizure severity. All data are presented asmeans ± the standard error of the mean; p values of

(15) McLean, M. J., Schmutz, M., Pozza, M. F., and Wamil, A. W.(2005) Effects of rufinamide on sodium-dependent action potentialfiring and sodium currents of rodent central neurons. Epilepsia 46,375−375.(16) Niespodziany, I., Leclere, N., Vandenplas, C., Foerch, P., andWolff, C. (2013) Comparative study of lacosamide and classicalsodium channel blocking antiepileptic drugs on sodium channel slowinactivation. J. Neurosci. Res. 91, 436−443.(17) Whitaker, W. R., Clare, J. J., Powell, A. J., Chen, Y. H., Faull, R.L., and Emson, P. C. (2000) Distribution of voltage-gated sodiumchannel α-subunit and β-subunit mRNAs in human hippocampalformation, cortex, and cerebellum. J. Comp. Neurol. 422, 123−139.(18) Catterall, W. A., Kalume, F., and Oakley, J. C. (2010) NaV1.1channels and epilepsy. J. Physiol. 588, 1849−1859.(19) Escayg, A., MacDonald, B. T., Meisler, M. H., Baulac, S.,Huberfeld, G., An-Gourfinkel, I., Brice, A., LeGuern, E., Moulard, B.,Chaigne, D., Buresi, C., and Malafosse, A. (2000) Mutations ofSCN1A, encoding a neuronal sodium channel, in two families withGEFS+2. Nat. Genet. 24, 343−345.(20) Papale, L. A., Beyer, B., Jones, J. M., Sharkey, L. M., Tufik, S.,Epstein, M., Letts, V. A., Meisler, M. H., Frankel, W. N., and Escayg, A.(2009) Heterozygous mutations of the voltage-gated sodium channelSCN8A are associated with spike-wave discharges and absenceepilepsy in mice. Hum. Mol. Genet. 18, 1633−1641.(21) Veeramah, K. R., O’Brien, J. E., Meisler, M. H., Cheng, X., Dib-Hajj, S. D., Waxman, S. G., Talwar, D., Girirajan, S., Eichler, E. E.,Restifo, L. L., Erickson, R. P., and Hammer, M. F. (2012) De novopathogenic SCN8A mutation identified by whole-genome sequencingof a family quartet affected by infantile epileptic encephalopathy andSUDEP. Am. J. Hum. Genet. 90, 502−510.(22) Hains, B. C., and Waxman, S. G. (2007) Sodium channelexpression and the molecular pathophysiology of pain after SCI. Prog.Brain Res. 161, 195−203.(23) Estacion, M., Gasser, A., Dib-Hajj, S. D., and Waxman, S. G.(2010) A sodium channel mutation linked to epilepsy increases rampand persistent current of Nav1.3 and induces hyperexcitability inhippocampal neurons. Exp. Neurol. 224, 362−368.(24) Vanoye, C. G., Gurnett, C. A., Holland, K. D., George, A. L., Jr.,and Kearney, J. A. (2014) Novel SCN3A variants associated with focalepilepsy in children. Neurobiol. Dis. 62, 313−322.(25) O’Kelly, I., Butler, M. H., Zilberberg, N., and Goldstein, S. A.(2002) Forward transport. 14-3-3 binding overcomes retention inendoplasmic reticulum by dibasic signals. Cell 111, 577−588.(26) Boileau, A. J., Kucken, A. M., Evers, A. R., and Czajkowski, C.(1998) Molecular dissection of benzodiazepine binding and allostericcoupling using chimeric γ-aminobutyric acidA receptor subunits. Mol.Pharmacol. 53, 295−303.(27) Trudeau, M. C., Warmke, J. W., Ganetzky, B., and Robertson, G.A. (1995) HERG, a human inward rectifier in the voltage-gatedpotassium channel family. Science 269, 92−95.(28) Dutton, S. B., Makinson, C. D., Papale, L. A., Shankar, A.,Balakrishnan, B., Nakazawa, K., and Escayg, A. (2012) Preferentialinactivation of Scn1a in parvalbumin interneurons increases seizuresusceptibility. Neurobiol. Dis. 49C, 211−220.(29) Burbidge, S. A., Dale, T. J., Powell, A. J., Whitaker, W. R., Xie, X.M., Romanos, M. A., and Clare, J. J. (2002) Molecular cloning,distribution and functional analysis of the NA(V)1.6. Voltage-gatedsodium channel from human brain. Brain Research. Molecular BrainResearch 103, 80−90.(30) Chen, Y. H., Dale, T. J., Romanos, M. A., Whitaker, W. R., Xie,X. M., and Clare, J. J. (2000) Cloning, distribution and functionalanalysis of the type III sodium channel from human brain. Eur. J.Neurosci. 12, 4281−4289.(31) Mantegazza, M., Yu, F. H., Powell, A. J., Clare, J. J., Catterall, W.A., and Scheuer, T. (2005) Molecular determinants for modulation ofpersistent sodium current by G-protein betagamma subunits. J.Neurosci. 25, 3341−3349.(32) Xie, X., Dale, T. J., John, V. H., Cater, H. L., Peakman, T. C., andClare, J. J. (2001) Electrophysiological and pharmacological properties

of the human brain type IIA Na+ channel expressed in a stablemammalian cell line. Pfluegers Arch. 441, 425−433.(33) Suter, M. R., Kirschmann, G., Laedermann, C. J., Abriel, H., andDecosterd, I. (2013) Rufinamide Attenuates Mechanical Allodynia in aModel of Neuropathic Pain in the Mouse and Stabilizes Voltage-gatedSodium Channel Inactivated State. Anesthesiology 118, 160−172.(34) Ledwell, J. L., and Aldrich, R. W. (1999) Mutations in the S4region isolate the final voltage-dependent cooperative step inpotassium channel activation. J. Gen. Physiol. 113, 389−414.(35) Kwan, P., Sills, G. J., and Brodie, M. J. (2001) The mechanismsof action of commonly used antiepileptic drugs. Pharmacol. Ther. 90,21−34.(36) Lipkind, G. M., and Fozzard, H. A. (2010) Molecular model ofanticonvulsant drug binding to the voltage-gated sodium channel innerpore. Mol. Pharmacol. 78, 631−638.(37) Rogawski, M. A. (2006) Diverse mechanisms of antiepilepticdrugs in the development pipeline. Epilepsy Res. 69, 273−294.(38) Chao, T. I., and Alzheimer, C. (1995) Effects of phenytoin onthe persistent Na+ current of mammalian CNS neurones. NeuroReport6, 1778−1780.(39) Arroyo, S. (2007) Rufinamide. Neurotherapeutics 4, 155−162.(40) Mohler, H. (2006) GABA(A) receptor diversity andpharmacology. Cell Tissue Res. 326, 505−516.(41) Trimmer, J. S., and Rhodes, K. J. (2004) Localization of voltage-gated ion channels in mammalian brain. Annu. Rev. Physiol. 66, 477−519.(42) Sievers, F., Wilm, A., Dineen, D., Gibson, T. J., Karplus, K., Li,W., Lopez, R., McWilliam, H., Remmert, M., Soding, J., Thompson, J.D., and Higgins, D. G. (2011) Fast, scalable generation of high-qualityprotein multiple sequence alignments using Clustal Omega. Mol. Syst.Biol. 7, 539.(43) Applegate, C. D., Samoriski, G. M., and Ozduman, K. (1997)Effects of valproate, phenytoin, and MK-801 in a novel model ofepileptogenesis. Epilepsia 38, 631−636.(44) Barton, M. E., Klein, B. D., Wolf, H. H., and White, H. S. (2001)Pharmacological characterization of the 6 Hz psychomotor seizuremodel of partial epilepsy. Epilepsy Res. 47, 217−227.(45) Ahmad, S., Dahllund, L., Eriksson, A. B., Hellgren, D., Karlsson,U., Lund, P. E., Meijer, I. A., Meury, L., Mills, T., Moody, A.,Morinville, A., Morten, J., O’Donnell, D., Raynoschek, C., Salter, H.,Rouleau, G. A., and Krupp, J. J. (2007) A stop codon mutation inSCN9A causes lack of pain sensation. Hum. Mol. Genet. 16, 2114−2121.(46) Singh, N. A., Pappas, C., Dahle, E. J., Claes, L. R., Pruess, T. H.,De Jonghe, P., Thompson, J., Dixon, M., Gurnett, C., Peiffer, A.,White, H. S., Filloux, F., and Leppert, M. F. (2009) A role of SCN9Ain human epilepsies, as a cause of febrile seizures and as a potentialmodifier of Dravet syndrome. PLoS Genet. 5, e1000649.(47) Glauser, T., Kluger, G., Sachdeo, R., Krauss, G., Perdomo, C.,and Arroyo, S. (2008) Rufinamide for generalized seizures associatedwith Lennox-Gastaut syndrome. Neurology 70, 1950−1958.(48) Hancock, E. C., and Cross, H. H. (2009) Treatment of Lennox-Gastaut syndrome. Cochrane Database of Systematic Reviews,CD003277.(49) Selmer, K. K., Lund, C., Brandal, K., Undlien, D. E., andBrodtkorb, E. (2009) SCN1A mutation screening in adult patientswith Lennox-Gastaut syndrome features. Epilepsy & Behavior 16, 555−557.(50) Claes, L. R., Deprez, L., Suls, A., Baets, J., Smets, K., Van Dyck,T., Deconinck, T., Jordanova, A., and De Jonghe, P. (2009) TheSCN1A variant database: A novel research and diagnostic tool. Hum.Mutat. 30, E904−E920.(51) Spampanato, J., Kearney, J. A., de Haan, G., McEwen, D. P.,Escayg, A., Aradi, I., MacDonald, B. T., Levin, S. I., Soltesz, I., Benna,P., Montalenti, E., Isom, L. L., Goldin, A. L., and Meisler, M. H.(2004) A novel epilepsy mutation in the sodium channel SCN1Aidentifies a cytoplasmic domain for β subunit interaction. J. Neurosci.24, 10022−10034.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121211

(52) Volkers, L., Kahlig, K. M., Verbeek, N. E., Das, J. H., vanKempen, M. J., Stroink, H., Augustijn, P., van Nieuwenhuizen, O.,Lindhout, D., George, A. L., Jr., Koeleman, B. P., and Rook, M. B.(2011) Nav 1.1 dysfunction in genetic epilepsy with febrile seizures-plus or Dravet syndrome. Eur. J. Neurosci. 34, 1268−1275.(53) Lossin, C. (2009) A catalog of SCN1A variants. Brain Dev. 31,114−130.(54) Dichgans, M., Freilinger, T., Eckstein, G., Babini, E., Lorenz-Depiereux, B., Biskup, S., Ferrari, M. D., Herzog, J., van denMaagdenberg, A. M., Pusch, M., and Strom, T. M. (2005) Mutationin the neuronal voltage-gated sodium channel SCN1A in familialhemiplegic migraine. Lancet 366, 371−377.(55) Cestele, S., Labate, A., Rusconi, R., Tarantino, P., Mumoli, L.,Franceschetti, S., Annesi, G., Mantegazza, M., and Gambardella, A.(2013) Divergent effects of the T1174S SCN1A mutation associatedwith seizures and hemiplegic migraine. Epilepsia 54, 927−935.(56) Moran, M., and Woiwode, T. (2009) Rufinamide for theTreatment of Post-Traumatic Stress Disorder. U.S. Patent20090054414 A1.(57) Oakley, J. C., Kalume, F., Yu, F. H., Scheuer, T., and Catterall,W. A. (2009) Temperature- and age-dependent seizures in a mousemodel of severe myoclonic epilepsy in infancy. Proc. Natl. Acad. Sci.U.S.A. 106, 3994−3999.(58) Yu, F. H., Mantegazza, M., Westenbroek, R. E., Robbins, C. A.,Kalume, F., Burton, K. A., Spain, W. J., McKnight, G. S., Scheuer, T.,and Catterall, W. A. (2006) Reduced sodium current in GABAergicinterneurons in a mouse model of severe myoclonic epilepsy ininfancy. Nat. Neurosci. 9, 1142−1149.(59) Han, S., Tai, C., Westenbroek, R. E., Yu, F. H., Cheah, C. S.,Potter, G. B., Rubenstein, J. L., Scheuer, T., de la Iglesia, H. O., andCatterall, W. A. (2012) Autistic-like behaviour in Scn1a± mice andrescue by enhanced GABA-mediated neurotransmission. Nature 489,385−390.(60) Weiss, L. A., Escayg, A., Kearney, J. A., Trudeau, M.,MacDonald, B. T., Mori, M., Reichert, J., Buxbaum, J. D., andMeisler, M. H. (2003) Sodium channels SCN1A, SCN2A and SCN3Ain familial autism. Mol. Psychiatry 8, 186−194.(61) Rogawski, M. A., and Loscher, W. (2004) The neurobiology ofantiepileptic drugs. Nat. Rev. Neurosci. 5, 553−564.(62) Rowley, N. M., and White, H. S. (2010) Comparativeanticonvulsant efficacy in the corneal kindled mouse model of partialepilepsy: Correlation with other seizure and epilepsy models. EpilepsyRes. 92, 163−169.(63) Stohr, T., Kupferberg, H. J., Stables, J. P., Choi, D., Harris, R. H.,Kohn, H., Walton, N., and White, H. S. (2007) Lacosamide, a novelanti-convulsant drug, shows efficacy with a wide safety margin inrodent models for epilepsy. Epilepsy Res. 74, 147−154.

ACS Chemical Biology Articles

dx.doi.org/10.1021/cb500108p | ACS Chem. Biol. 2014, 9, 1204−12121212